Hello LAM TEAM/USERS,
I have gone through the following links .
http://gridengine.sunsource.net/howto/lam-integration/lam-integration.html
There I followed the loose integration of LAM with SGE
using rsh. I downloaded the sge-lam script and copied
it to the SGE_ROOT/mpi directory. In sge-lam directory
i found two scripts (a) startlam.sh (b) stoplam.sh
I did not touch those two scripts.
I added one pe named loose_lam_rsh in parallel
environment of QMON. There for "Start Proc Arg" field
I gave the path to startlam.sh and for "Stop Proc Arg"
field I gave the path to stoplam.sh. Then I run my job
through following command
qsub -pe loose_lam_rsh 4 job
Job is pending. It is not executed.
My Query:
(1): Do I need to change anything in startlam.sh or
stoplam.sh?
(2): In startlam.sh script i found some /machines, but
previously I have created "hostfile" which will have
all node names and no of cpus and that "hostfile" i
have put it in all the nodes at same path. I am able
to run successfully "lamboot -d hostfile"
I am getting all my nodes in "lamnodes" command.But
still what is again to create /machines? Do i need to
do anything else after downloading sge-lam script.
(3) Can you please provide me a consize process to
integrate sge-lam ( I am bit confused if i go through
the above link) So please give me consize
instructions.
Through mpirun my job is running fine only i need to
run it through qsub so plz give me in details about
Integrating lam with sge.
I am using lam 7.1.1 on my Bio Clusture grid. OS is
RedHat WS 3.0 for AMD 64 Architecture.Machine type sun
fire v20z.
I need your earlier assitance.
With Thanks and regards
Debasis
__________________________________
Yahoo! Mail Mobile
Take Yahoo! Mail with you! Check email on your mobile phone.
http://mobile.yahoo.com/learn/mail
|
